Review Reports - The Effects of Interest Rates on Bank Risk-Taking in South Africa: Do Cyclical and Location Asymmetries Matter?

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

I believe that authors have address an interesting an challenging topic. The paper contains important information on existing studies and provides a comparison between several methods with specific data set. I would recommend the authors to consider the following minor improvements:

- extend a little bit the conclusion, as there are recommendations to reassess macroprudential policies, and include more detailed suggestions.

- provide more specific reference to the graphics and comparison of the series presented at Figure 1.

- literature review can be extended to provide information on studies covering other (not just post 2008 crisis) periods.

Author Response

The authors would like to thank the reviewer for the comments made. We believe they have improved our paper. See below our response to the comments made.

Reviewer 1

Comment 1: extend a little bit the conclusion, as there are recommendations to reassess macroprudential policies, and include more detailed suggestions.

Response: A recommendation on macroprudential regulation has been strengthened in the conclusion and recommendations section.

Comment 2: provide more specific reference to the graphics and comparison of the series presented at Figure 1.

Response: Reference to figure 1 has been included in the introduction of the paper. This also serves as a motivation for the selectin of the estimation techniques.

Comment 3: literature review can be extended to provide information on studies covering other (not just post 2008 crisis) periods.

Response: The impact of the interest rate on bank risk taking gained traction post the 2008/09 global financial crisis, hence the authors reviewed studies post the crisis. We have found no literature on the topic prior to the crisis. Moreover, we have exhausted all available studies in the literature

Reviewer 2 Report

Comments and Suggestions for Authors

ijfs-2922668

I would appreciate the editor for providing me an opportunity to review the article “The effects of interest rates on bank risk-taking in South Africa: Do cyclical and location asymmetries matter?”. In this paper, the authors examine the nonlinear relationship between interest rates on bank risk-taking behavior in South Africa. The paper is overall interesting, but I still have some concerns on this paper.

1. The research motivation should be clarified. Why do the authors focus on cyclical and location asymmetries? Is there any practical motivation of conducting such a study? The authors may clarify the motivation in Introduction to help the readers understand the importance of this paper.

2. I am concerned of how NARDL can account for the influence of cyclical. As shown in Subsection 3.2, only the lagged term of the explanatory and control variables are included in the model, and they are unrelated to cyclical in my opinion.

3. What is the motivation of employing QARDL in this paper? Please provide some explanations.

4. The different signs and significance of INT on the dependent variable may be hard to interpret. The results are not robust in my opinion.

5. Table 8 reports the results of QARDL, not NARDL.

6. The authors seem to fail to explain the location asymmetries as shown in Table 7.

7. I have a feeling that the authors focus too much on the methodology rather than the economic meaning. Maybe the authors can explore the topic via a panel data, which will definitely provide some new evidence.

Author Response

The authors would like to thank the reviewer for the comments made. We believe they have improved our paper. See below our response to the comments made.

Reviewer 2

Comment 1: The research motivation should be clarified. Why do the authors focus on cyclical and location asymmetries? Is there any practical motivation of conducting such a study? The authors may clarify the motivation in Introduction to help the readers understand the importance of this paper.

Response: We provide a motivation for our use of location and cyclical asymmetries in the 4^th paragraph

Comment 2. I am concerned of how NARDL can account for the influence of cyclical. As shown in Subsection 3.2, only the lagged term of the explanatory and control variables are included in the model, and they are unrelated to cyclical in my opinion.

Response: The NARDL is a model which partitions the data into cyclical components. If you pay attention to equation (6) and (7) you should notice that we segregate the effects of increasing interest rate () from decreasing interest rate ) and this induces cyclical variation in the regression estimates.

Comment 3. What is the motivation of employing QARDL in this paper? Please provide some explanations.

Response: We motivate the use of the QARDL model to location asymmetries. We provide an improved explanation of this in the 4^th paragraph in the introduction

Comment 4. The different signs and significance of INT on the dependent variable may be hard to interpret. The results are not robust in my opinion.

Response: This comment is a bit difficult to understand. The coefficient on the interest rate variable is expected to produce different signs on the variables since different measures of risk taking our used in our study. We consider the results robust as they are the

Comment 5. Table 8 reports the results of QARDL, not NARDL.

Response: We have corrected this

Comment 6: The authors seem to fail to explain the location asymmetries as shown in Table 7.

Response: Contrary to this comment, location asymmetries are discussed under the QARDL models reported in Tables 8 and 9. Table 7 discusses NARDL reports cyclical asymmetries from the NARDL model.

Comment 7: I have a feeling that the authors focus too much on the methodology rather than the economic meaning. Maybe the authors can explore the topic via a panel data, which will definitely provide some new evidence.

Response: We reserve the suggestion of extending the current study to a panel study for future research.

Reviewer 3 Report

Comments and Suggestions for Authors

This paper utilizes South African banks as a laboratory and employs nonlinear estimation techniques to study how interest rate changes affect bank risk-taking activities in a non-symmetric way. The authors conclude that the relationship between interest rates and banks’ interest-taking behaviors are asymmetrically affected by both different monetary policies and different levels of risk environment. The paper can be further improved in the following areas.

The first is motivation. The authors demonstrate that two factors (i.e., monetary policy and location) have complicated effects on the relationship under concern. The authors need first to restate how the interest rate is related to bank risk-taking under the assumption that monetary policy and location variables are held constant or at an average level and then clarify why, in South Africa, the above two variables are essential and selected as moderating variables on the above-concerned relation (might be there are deviations from the conventional theory prediction that points to them as underlying reasons).

The second concerns methodology. Nonlinear estimation methods also include threshold effects and higher moments. Why is asymmetry the focus? Why are QARDL and NARDL preferred over other alternatives? The authors need to state the advantages and disadvantages and explain how they adjust those methods to fit the current setup.

Third, sample. The authors must clearly state the reason for choosing South Africa as a laboratory to study this research question.

Fourth, robustness. More robustness tests are needed to persuade authors that the findings are valid. In addition, the interpretations of the results need to be more convincing.

Finally, there are also some minor issues. For example, an institutional background or stylized facts should be added to the literature review before reviewing theoretical and empirical papers. For another example, the conclusion has to be expanded by including the limitations of the present study and the directions of future studies. For the last example, the tables do not follow the academic standard reporting format, and some tables can really go to the appendix, such as the variable explanation and correlation matrix. It would also be better if some charts (on top of Figure 1) could be added to illustrate some points on asymmetry.

Comments on the Quality of English Language

English can be more concise and use less passive tone.

Author Response

The authors would like to thank the reviewer for the comments made. We believe they have improved our paper. See below our response to the comments made.

Reviewer 3

Comment 1: The first is motivation. The authors demonstrate that two factors (i.e., monetary policy and location) have complicated effects on the relationship under concern. The authors need first to restate how the interest rate is related to bank risk-taking under the assumption that monetary policy and location variables are held constant or at an average level and then clarify why, in South Africa, the above two variables are essential and selected as moderating variables on the above-concerned relation (might be there are deviations from the conventional theory prediction that points to them as underlying reasons).

Response: We have tried to improve on the motivation of the study in the introduction.

Comment 2: The second concerns methodology. Nonlinear estimation methods also include threshold effects and higher moments. Why is asymmetry the focus? Why are QARDL and NARDL preferred over other alternatives? The authors need to state the advantages and disadvantages and explain how they adjust those methods to fit the current setup.

Response: In the introduction, we have included ‘new’ fourth and fifth paragraphs of the introduction in which motivate the use of the NARDL and QRADL models.

Comment 3: Third, sample. The authors must clearly state the reason for choosing South Africa as a laboratory to study this research question.

Response: We motivated the use of South Africa as our focus country in the first two paragraphs of the introduction section.

Comment 4: Fourth, robustness. More robustness tests are needed to persuade authors that the findings are valid. In addition, the interpretations of the results need to be more convincing.

Response: We use three different estimators to ensure robustness of our empirical exercise. We also use three different measures of bank risk taking to further ensure robustness.

Comment 5: Finally, there are also some minor issues. For example, an institutional background or stylized facts should be added to the literature review before reviewing theoretical and empirical papers. For another example, the conclusion has to be expanded by including the limitations of the present study and the directions of future studies. For the last example, the tables do not follow the academic standard reporting format, and some tables can really go to the appendix, such as the variable explanation and correlation matrix. It would also be better if some charts (on top of Figure 1) could be added to illustrate some points on asymmetry.

Response: Figure 1 has been added to the introduction section. The authors felt that this is the best place to discuss the trends in the variables as this provides motivation for the selection of the chosen techniques. Areas of future research and limitations/delimitations have been added at the end of the conclusion and recommendations section.

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

I have no more comments on this paper now except that Table 2 can be further well-formatted.

Author Response

The authors tried different options for formatting Table 2. However, this turned out to be the best format given the information that we wanted to highlight.

Reviewer 3 Report

Comments and Suggestions for Authors

Thank you for addeessing my previous comments. The paper was improved a lot and can be better by considering the following minor problems. First, table 1 should be updated to most recent literature and working paper, and should add a new column to compare the literature to this paper. Second, the methodology part can be shorterned. Third, the paper needs to talk about a little bit about results robustness and the underlying mechanism. Finally, table 8-9 should be more consistent in format. e.g., why the lend and loans variable have different ranks? * is not explained in the notes. short and long run results and significance for some variables change fundamentally, which needs more explanation.

Comments on the Quality of English Language

English fine

Author Response

Comment 1: We have already added all the literature. We have done Google Scholar searches and have not found additional literature. Hence this study is an important one. The reviewer also wants us to compare the findings of our current study to that of the literature in the Tables presented in the literature review section. However, we felt that it is better to make comparisons with previous literature in the results section rather than in the literature section. In the results section, we show how our study is similar to or differs from the literature.

Comment 2: The methodology section has been written as concisely as possible. We cannot shorten this section without removing vital information from the models.

Comment 3: The robustness of the results lies in the fact that we have used three different estimation techniques. The results are roughly similar for all the models. Furthermore, we used three different measures of risk taking which should also add to the robustness of the results.

Comment 4: Table 9 has been changed such that there is consistency in the variables. There is no rank between the dependent variables. They are equally important. Also, we have explained that the long-run and short-run coefficients generally mirror each but differ a bit in significance (with long-run generally having more significant estimates than short-run coefficients). The general explanation is that the relationships are 'more solid' over the long-run than short run. Furthermore, the results suggest that banks alter their lending behaviour in response to macroeconomic variables in the long-run compared to the short-run mostly in response to changes in profit levels.